CN104537438A - Forecast and monitoring method for peak-hour power usage - Google Patents

Abstract

The invention discloses a method for monitoring electricity consumption peaks. The method includes: classifying electricity consumption data according to preset temperature thresholds, and determining the category of electricity consumption data in a time period to be predicted; Collect the power consumption data of the specified category of all electrical equipment to be tested in the section; establish a conditional random field model, and use the collected power consumption data as training samples to estimate the conditional probability distribution function of power consumption; according to the The conditional probability distribution function of power consumption and the current power consumption predict the expected power consumption in the next forecast period and whether there will be a peak power consumption; when it is predicted that there will be a peak power consumption, an alarm message will be sent to the user. By using the above method, the power consumption of the power grid can be well monitored, so as to adjust the power of electrical appliances in time to alleviate the load of the power grid before the peak of power consumption occurs.

Description

A Method for Predicting and Monitoring Power Consumption Peak

technical field

The invention relates to the technical field of power grid load control, in particular to a method for forecasting and monitoring peak power consumption.

Background technique

With the continuous improvement of living standards, the proportion of residential electricity consumption in the total load of the grid is increasing. Due to the time-concentrated characteristics of residential electricity consumption, it will cause the grid load to fluctuate sharply in a short period of time, resulting in peak electricity consumption. The measures taken by the traditional power system to cope with peak loads are mainly achieved by increasing the installed capacity of power generation and improving the power transmission and distribution capacity of power grid equipment, which makes the utilization efficiency of power generation side and power grid side equipment low and seriously wastes resources.

It can be seen that power grid load forecasting is of great significance to power grid load regulation. At present, the mainstream power grid load forecasting methods used in the prior art include: regression analysis forecasting method (including linear regression and nonlinear regression methods), time series forecasting method, gray forecasting method, neural network forecasting method and so on.

However, the grid load forecasting methods in the prior art (for example, take the regression method as an example) generally have disadvantages such as huge amount of calculation and poor real-time performance, so it is difficult to better predict the grid load of residential users.

Contents of the invention

In view of this, the purpose of the present invention is to propose a method for forecasting and monitoring peak power consumption, so that the grid load of residential users can be better monitored, so as to adjust the power of electrical appliances in time before the peak power consumption occurs to alleviate the grid load. load.

Based on the above purpose, the present invention provides a method for monitoring peak power consumption, the method comprising:

Classify the power consumption data according to the preset temperature threshold, and determine the category of the power consumption data within the time period to be predicted;

Collect the electricity consumption data of the specified category of all electrical equipment to be tested within the time period to be predicted;

Establish a conditional random field model, and use the collected electricity consumption data as training samples to estimate the conditional probability distribution function of electricity consumption;

According to the conditional probability distribution function of the power consumption and the current power consumption, predict the expected power consumption in the next forecast period and whether there will be a peak power consumption;

When it is predicted that there will be a peak power consumption, an alarm message will be sent to the user.

Preferably, the conditional probability distribution function of electricity consumption obtained by estimating the collected electricity consumption data as training samples includes:

Initialize the conditional random field model;

Input the collected power consumption data as training samples into the conditional random field model after initialization for iterative calculation, and use the maximum likelihood parameter estimation algorithm to estimate the value of the characteristic weight parameter λ, so as to obtain the power consumption Conditional probability distribution function.

Preferably, the conditional probability distribution function of electricity consumption is calculated using cloud computing technology.

Preferably, the conditional probability distribution function of the electricity consumption is:

$\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; \&lambda;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; Z</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>$

Among them, p(y|x,λ) is the conditional probability distribution function of electricity consumption, x is the current electricity consumption, y is the expected electricity consumption, λ is the characteristic weight parameter, Z(x) is the normalization factor, f is a feature vector.

Preferably, the initial setting of the conditional random field model includes:

Set the initial value of the feature weight parameter λ to 0.

Preferably, said predicting whether there will be a power consumption peak in the next forecast period includes:

Preset the power consumption threshold P _a and the probability threshold P _t of the peak power consumption;

Calculate the probability that the value of the expected power consumption is greater than P _a according to the conditional probability distribution function of the power consumption;

When the probability that the value of the expected power consumption is greater than P _a is greater than or equal to the probability threshold P _t of the peak power consumption, it is judged that there will be a peak power consumption in the next forecast period.

As can be seen from the above, because in the monitoring method of the peak power consumption in the present invention, the conditional random field theory is used to establish a mathematical model for a large number of household user electrical equipment, and predict the total power consumption of residents with this By predicting the peak of electricity consumption and monitoring the peak of electricity consumption, measures can be taken in advance to reasonably adjust the use of electricity to achieve the purpose of reducing the peak and balancing the relationship between power supply and demand, so that the automatic control technology of smart home can be used in Before the expected electricity consumption peak occurs, adjust the electrical power in time to achieve the purpose of alleviating the grid load. Therefore, it has higher accuracy and real-time performance than other prediction and regulation methods in the prior art. Combined with smart home technology, it can Quickly and effectively adjust the grid load when the existing grid operation is affected.

Description of drawings

FIG. 1 is a schematic flow diagram of a method for monitoring peak power consumption in an embodiment of the present invention;

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be described in further detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

This embodiment provides a method for monitoring peak power consumption.

FIG. 1 is a schematic flowchart of a method for monitoring peak power usage in an embodiment of the present invention. As shown in Figure 1, the monitoring method of the peak power consumption in the embodiment of the present invention mainly includes:

Step 11, classify the power consumption data according to the preset temperature threshold, and determine the category of the power consumption data within the time period to be predicted.

In the technical solution of the present invention, weather factors, such as the influence of temperature on power consumption, need to be considered, and the power consumption data under each type of specific weather condition has a corresponding rule. Therefore, a temperature threshold can be preset, and then the power consumption data can be classified according to the preset temperature threshold. For example, in a preferred embodiment of the present invention, the electricity consumption data under four conditions of cold weather, cool weather, warm weather and hot weather can be classified.

After the power consumption data is classified, the category of the power consumption data in the time period to be predicted can be determined according to the temperature value in the time period to be predicted.

Step 12, collect electricity consumption data of a specified category of all electrical equipment to be tested within the time period to be predicted.

Preferably, in a specific embodiment of the present invention, the smart home system can be combined with the Internet of Things technology to collect the power consumption data of the specified category of all the electrical equipment to be tested in the system within the time period to be predicted, and the collected The collected power consumption data is summarized into the database of the server, so that the collected power consumption data can be used as training samples in the subsequent step 13.

Step 13, establishing a conditional random field model, and using the collected power consumption data as training samples to obtain a conditional probability distribution function of power consumption.

Conditional Random Fields (CRF) theory is essentially a statistical learning method. Statistical learning constructs probabilistic and statistical models by analyzing large amounts of data, extracts the characteristics of the data, and makes predictions on the trend of the data.

The conditional random field model is a statistical model used to label and slice serialized data. These data are preset to have Markov properties. The model computes the joint probability of the entire labeled sequence given the sequence of observations to be labeled. The conditional properties of the distribution of the marker sequence can make the conditional random field fit the real data well, and in these data, the conditional probability of the marker sequence depends on the non-independent and interacting features in the observed sequence, and by endowing the features with Different weights represent the importance of features.

The conditional random field model is a model established for a group of random variables with Markov properties. The Markov property means that random variables based on undirected graph topological union are only related to adjacent variables, but independent of non-adjacent variables.

In the technical solution of the present invention, the power consumption of each measured electric device is used as a variable of the conditional random field model, so that the conditional random field model can be established, and according to the established conditional random field model, conditional random field Theory and estimation methods to forecast electricity consumption trends.

Preferably, in a specific embodiment of the present invention, the conditional probability distribution function of the electricity consumption can be expressed as:

$\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; \&lambda;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; Z</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>$

Among them, p(y|x,λ) is the conditional probability distribution function of electricity consumption, x is the current electricity consumption, y is the expected electricity consumption, λ is the characteristic weight parameter, Z(x) is the normalization factor, f is a feature vector.

According to the above formula, when the conditional random field model is initially established, the characteristic weight parameter λ in the conditional probability distribution function of electricity consumption in the above conditional random field model is an unknown parameter (that is, a parameter whose value is unknown). Therefore, in the technical solution of the present invention, the collected electricity consumption data can be used as training samples to estimate the value of the above-mentioned characteristic weight parameter λ, thereby obtaining the conditional probability distribution function of electricity consumption.

For example, preferably, in a preferred embodiment of the present invention, the conditional probability distribution function of electricity consumption obtained by estimating the collected electricity consumption data as training samples can be realized through the following steps:

Step 131, initialize the conditional random field model.

For example, preferably, in a specific embodiment of the present invention, the initial setting of the conditional random field model includes: setting the initial value of the feature weight parameter λ to 0.

Of course, in the technical solution of the present invention, the initial value of the feature weight parameter λ can also be set to other values according to the needs of practical applications.

Step 132, input the collected power consumption data as training samples into the conditional random field model after initialization for iterative calculation, and use the maximum likelihood parameter estimation algorithm to estimate the value of the characteristic weight parameter λ, so as to obtain The conditional probability distribution function of the electric quantity.

In the technical solution of the present invention, since the collected power consumption data is large and the calculation is complicated, it is difficult for a general computer to meet the above calculation requirements. Therefore, preferably, in a specific embodiment of the present invention, you can use Cloud computing technology calculates the conditional probability distribution function of electricity consumption.

Step 14: According to the conditional probability distribution function of the electricity consumption and the current electricity consumption, predict the expected electricity consumption in the next forecast period and whether there will be a peak in electricity consumption.

Since the conditional probability distribution function of electricity consumption is obtained in step 13, in this step, the expected electricity consumption in the next forecast period can be calculated according to the conditional probability distribution function of the electricity consumption and the current electricity consumption Forecasting is made and it is also possible to predict whether there will be a peak in electricity usage during the next forecast period.

For example, preferably, in a specific embodiment of the present invention, said predicting whether there will be a power consumption peak in the next forecast period includes:

Step 141, preset the power consumption threshold P _a and the probability threshold P _t of the peak power consumption.

Step 142, calculating the probability that the value of the expected power consumption is greater than P _a according to the conditional probability distribution function of the power consumption.

Step 143, when the probability that the value of the expected power consumption is greater than P _a is greater than or equal to the probability threshold P _t of the peak power consumption (that is, p(y|x>P _a )≥P _t ), determine the next forecast time There will be a peak in power consumption during the period.

Step 15, when it is predicted that there will be a power consumption peak, send an alarm message to the user, so that the user can intervene in the user's power consumption behavior according to the preset power consumption strategy, reduce power consumption, and avoid the power consumption peak.

For example, in a preferred embodiment of the present invention, the intervening on the user's electricity consumption behavior may include: controlling the operating large energy-consuming electrical appliances (eg, water heaters, etc.) to reduce operating power, or even shut down.

Through the above steps 11-15, the monitoring of the peak power consumption can be realized.

In summary, because in the monitoring method of electricity consumption peak in the present invention, the conditional random field theory is used to establish a mathematical model for a large number of household electrical equipment, and to predict the trend of the total electricity consumption of residents, through Predict the time period of the peak power consumption, and take measures in advance to adjust the power consumption reasonably, so as to achieve the purpose of reducing the peak value and balancing the relationship between power supply and demand, so that the automatic control technology of smart home can be used to adjust the power of electrical appliances in time before the expected peak power consumption occurs In order to achieve the purpose of alleviating the grid load, it has higher accuracy and real-time performance than other forecasting and regulation methods in the existing technology. Regulate grid load.

Those of ordinary skill in the art should understand that: the above descriptions are only specific embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present invention etc., should be included within the protection scope of the present invention.

Claims (6)

1. A monitoring method for peak power consumption, characterized in that the method comprises Classify the power consumption data according to the preset temperature threshold, and determine the category of the power consumption data within the time period to be predicted; Collect the electricity consumption data of the specified category of all electrical equipment to be tested within the time period to be predicted; Establish a conditional random field model, and use the collected electricity consumption data as training samples to estimate the conditional probability distribution function of electricity consumption; According to the conditional probability distribution function of the power consumption and the current power consumption, predict the expected power consumption in the next forecast period and whether there will be a peak power consumption; When it is predicted that there will be a peak power consumption, an alarm message will be sent to the user. 2. The method according to claim 1, wherein the conditional probability distribution function of the electricity consumption obtained by estimating the collected electricity consumption data as a training sample comprises: Initialize the conditional random field model; Input the collected power consumption data as training samples into the conditional random field model after initialization for iterative calculation, and use the maximum likelihood parameter estimation algorithm to estimate the value of the characteristic weight parameter λ, so as to obtain the power consumption Conditional probability distribution function. 3. The method according to claim 2, characterized in that: The conditional probability distribution function of electricity consumption is calculated by using cloud computing technology. 4. The method according to claim 2, wherein the conditional probability distribution function of the power consumption is: $\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; \&lambda;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; Z</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>$ Among them, p(y|x,λ) is the conditional probability distribution function of electricity consumption, x is the current electricity consumption, y is the expected electricity consumption, λ is the characteristic weight parameter, Z(x) is the normalization factor, f is a feature vector. 5. method according to claim 4, is characterized in that, described conditional random field model is initialized and set comprises: Set the initial value of the feature weight parameter λ to 0. 6. The method according to claim 1, wherein the predicting whether a power consumption peak occurs in the next forecast period comprises: Preset the power consumption threshold P _a and the probability threshold P _t of the peak power consumption; Calculate the probability that the value of the expected power consumption is greater than P _a according to the conditional probability distribution function of the power consumption; When the probability that the value of the expected power consumption is greater than P _a is greater than or equal to the probability threshold P _t of the peak power consumption, it is judged that there will be a peak power consumption in the next forecast period.